[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: Is OO a deliberate fraud?

Quoting neil@xxxxxxxxxxxxxxxxx:

> As for webserves, this caught my eye recently:
> http://www.sics.se/~joe/apachevsyaws.html
> Threads are definitely too heavyweight!

I have to admit that I'm quite surprised that the CSP community hasn't embraced
Erlang more. Here's a language built entirely around the notion of processes
that communicate by message passing, with the internal behavior of each process
defined in a tail-recrusive functional style. From everything I've heard, the
concurrency available in Erlang is very low overhead, and Erlang programs often
contain huge numbers of processes.

About the only significant difference between the Erlang model and CSP is that
Erlang's message-passing is asynchronous rather than synchronous. But it's no
trouble at all to either emulate synchronous message-passing in Erlang, or
asynchronous message-passing in CSP. In fact, there's already a tool for
translating Erlang into the mCRL process algebra
(http://sourceforge.net/projects/etomcrl/), so it presumably wouldn't be too
hard to establish a similar mapping to CSP (and vice versa).

Is there some reason I'm not aware of that Erlang hasn't been a more popular
target for CSP-based development?

Allan McInnes <amcinnes@xxxxxxxxxx>
PhD Candidate
Dept. of Electrical and Computer Engineering
Utah State University